@ARTICLE{Sowińska-Botor_Justyna_Ranking_2023,
 author={Sowińska-Botor, Justyna and Mastej, Wojciech and Maćkowski, Tomasz},
 volume={vol. 39},
 number={No 3},
 journal={Gospodarka Surowcami Mineralnymi - Mineral Resources Management},
 pages={149-172},
 howpublished={online},
 year={2023},
 publisher={Komitet Zrównoważonej Gospodarki Surowcami Mineralnymi PAN},
 publisher={Instytut Gospodarki Surowcami Mineralnymi i Energią PAN},
 abstract={The suitability of several low-labor geostatistical procedures in the interpolation of highly positively skewed seismic data distributions was tested in the Baltic Basin. These procedures were a combination of various estimators of the model of spatial variation (theoretical variogram) and kriging techniques, together with the initial data transformation to normal distribution or lack thereof. This transformation consisted of logarithmization or normalization using the anamorphosis technique. Two variations of the theoretical variogram estimator were used: the commonly used classical Matheron estimator and the inverse covariance estimator (InvCov), which is robust with regard to non-ergodic data. It was expected that the latter would also be resistant to strongly skewed data distributions. The kriging techniques used included the commonly used ordinary kriging, simple kriging useful for standardized data and the non-linear median indicator kriging technique. It was confirmed that normalization (anamorphosis) is the most useful and less laborious geostatistical procedure of those suitable for such data, which results in a standardized normal distribution. The second, not obvious statement for highly skewed data distributions suggests that the non-ergodic inverted covariance (InvCov) estimator of variogram has an advantage over the Matheron’s estimator. It gives a better assessment of the C 0 (nugget effect) and C (sill) parameters of the spatial variability model. Such a conclusion can be drawn from the fact that the higher the estimation of the relative nugget effect L = C 0/(C 0 + C) using the InvCov estimator, the weaker the correlation between the kriging estimates and the observed values. The values of the coefficient L estimates obtained by using the Matheron’s estimator do not meet this expectation.},
 type={Article},
 title={Ranking of the utility of selected geostatistical interpolation methods in conditions of highly skewed seismic data distributions: a case study of the Baltic Basin (Poland)},
 URL={http://journals.pan.pl/Content/128679/PDF/Sowinska-Bator%20i%20inni.pdf},
 doi={10.24425/gsm.2023.147555},
 keywords={seismic, data processing, shallow subsurface, uncertainty, variability},
}